Python implementation of the Forecasting Inundation Extents using REOF method
Based off of the methods from Chang et al., 2020
$ conda create -n fier -c conda-forge python=3.8 netcdf4 qt pyqt rioxarray numpy scipy xarray pandas scikit-learn eofs geoglows
$ conda activate fier
$ pip install git+https://github.com/servir/fierpy.gitTo Install in OpenSARlab:
$ conda create --prefix /home/jovyan/.local/envs/fier python=3.8 netcdf4 qt pyqt rioxarray numpy scipy xarray pandas scikit-learn eofs geoglows jupyter kernda
$ conda activate fier
$ pip install git+https://github.com/servir/fierpy.git
$ /home/jovyan/.local/envs/fier/bin/python -m ipykernel install --user --name fier
$ conda run -n fier kernda /home/jovyan/.local/share/jupyter/kernels/fier/kernel.json --env-dir /home/jovyan/.local/envs/fier -o- numpy
- xarray
- pandas
- eofs
- geoglows
- scikit-learn
- rasterio
import xarray as xr
import fierpy
# read sentinel1 time series imagery
ds = xr.open_dataset("sentine1.nc")
# apply rotated eof process
reof_ds = fierpy.reof(ds.VV,n_modes=4)
# get streamflow data from GeoGLOWS
# select the days we have observations
lat,lon = 11.7122,104.9653
q = fierpy.get_streamflow(lat,lon)
q_sel = fierpy.match_dates(q,ds.time)
# apply polynomial to different modes to find best stats
fit_test = fierpy.find_fits(reof_ds,q_sel,ds)